Visualización de las propiedades del objeto de datos

The second hypothesis presented in chapter 6 was that levels of migration aspirations would be higher in areas where such networks were prevalent. I based this hypothesis on the culture of migration theory, which implies that in areas where prevalence of out-migration is high, aspiration levels will rise accordingly, as ideas of migration as a valuable life project are diffused throughout the community. The culture of migration theory and migrant networks theory are linked, but they differ on one important point. Cultures of migration are the prevalence of migrant networks in ones local community, while migrant networks are personal and related to individuals knowing other individuals personally. Cross-tabulations of areas and migration aspirations in table 7.3 yielded some support to the culture of migration theory and my second hypothesis, but table 7.3 gives us further indication of whether this is also a valid finding when we control for other determinants of migration aspirations.

The four areas surveyed by EUMAGINE were selected because the team wanted to have different emigration environments to compare, and the degree to which they differ in regards the prevalence of migrant networks and histories of migration was discussed in chapter 5. The descriptive statistics in table 7.1 indicated that there seems to be a relatively low prevalence of migrant networks in Lambaye and a relatively high prevalence of such networks in Orkadiéré. This was also the pattern found when measuring the prevalence of respondents who themselves have a personal history of international migration, presented in

73 chapter 5. 2.5 percent of the respondents in Lambaye have personal history of international migration, while over 10 percent of the respondents in Orkadiéré report of having lived abroad for a period of time²². According to Kandel and Massey’s (2002) definition of a culture of migration, this places Orkadiéré in the high prevalence category, and Lambaye in the low-prevalance, as was also discussed in chapter 5. So far we have only looked at this descriptively. The observed differences may therefore be caused by coincidence in respondent’s answers, a factor which will be tested in the logistic regression analysis.

In table 7.4, the areas are included only in model two and three. The area-coefficients for both Golf Sud and Orkadiéré are positive and significant, which means that people living in Golf Sud or Orkadiéré, have significantly higher levels of migration aspirations than people living in Lambaye, which is the reference category, controlled for other determinants of migration aspirations. These indications from the descriptive statistics are thus confirmed, and are not only valid for the respondents, but for the population at large. People from Orkadiéré are almost three times as likely to aspire to migrate, compared to respondents from Lambaye.

People from Golf Sud are over twice as likely, both controlled for other relevant determinants of migration aspirations. The coefficient for Darou Mousty is negative, but not statistically significant, so we can not say that there is a significant difference between people living in Lambaye and people living in Darou Mousty as regards their levels of migration aspirations.

Based on this findings, it is clear that there are differences in the levels of migration aspirations between the areas, and according to Kandel and Massey’s (2002) understanding of the term, this yields support of the culture of migration theory. Areas where prevalence of networks are high are also the areas where migration aspirations are high.

It is important to remember that while these results give reason to believe that there is a stronger, and more established culture of migration in Orkadiéré and Golf Sud than we find in Lambaye, the areas also differ on a number of other factors, on which EUMAGINE has not collected data. The lack of other data on these areas means that we have to interpret the results with caution, and be open for the possibility that there may be other area-specific factors that are causing the observed differences, which have not been possible to control for given the existing data structure and the lack of specific population data on the four areas. These un-observed factors may be causing both high aspiration level, and be what lead many to migrate in the past, which drives the prevalence of migrant networks up. With this reservation in

22 Whether or not the areas differ could also be found by performing a chi-test. Such tests have been performed, but are not reported here because the results from the logistic regression models provide the same information.

mind, and knowing that we have controlled for important determinants of migration aspirations in the model, I chose to conclude that the findings we see in table 7.4 are a confirmation of the second hypothesis, and an indication that people who are living in areas with high prevalence of migrant networks, are more likely to aspire to migrate.

The third hypothesis I presented in chapter 6 was that access to personal migrant networks would be less important for aspirations in areas where migrant networks are prevalent in the community, i.e. where a culture of migration is more strongly established. To test this assumption empirically, model three in table 7.4 provides interaction terms between the areas and the migrant network variables. When interpreting the coefficients in model three, it is important to keep in mind that adding interaction terms to a model changes the interpretation of all coefficients included in the interaction. Thus, network coefficients in model three indicate the odds for the reference category, which in this case is people from Lambaye. For example, the coefficient for knowing a returned migrant changes substantially from model two to model three, because in model two it indicated the effect of knowing returned migrants when all other factors were held constant, while in model three, the coefficient for knowing returned migrants only denotes the effect for respondents in Lambaye.

We see in model three in table 7.4, that there are two significant interaction terms, between returned migrants and living in Darou Mousty, and between returned migrants and living in Orkadiéré. Both are negative. The interaction coefficient between knowing a returned migrant and living in Orkadiéré shows the added effect of such networks for people in Orkadiéré, compared to residents living in Lambaye. It proves that there is significant difference between living in Orkadiéré or Darou Mousty and living in Lambaye for the degree to which migrant networks influence migration aspirations. Said simply, knowing returned migrant networks is less important for migration aspirations in Orkadiéré and Darou Mousty than in Lambaye, the area where a culture of migration is the least prevalent. It is important to keep in mind that since the aspiration levels were already substantially higher in Orkadiéré than in Lambaye, people with such networks in Lambaye do not necessarily have higher levels of migration aspirations than people in Orkadiéré without such networks. Rather, what these interaction terms indicate is that having migrant networks is more important for aspirations in Lambaye than it is in Orkadiéré and Golf Sud. This is seems to confirm also the third hypothesis, at least for people who know returned migrants. Knowing returned migrants is more important for aspirations in Lambaye than it is in Golf Sud or Orkadiéré, where such

75 networks are prevalent. According to the culture of migration theory, this might be because the idea of migration being a positive life project or an established strategy for upward social mobility, is already established in Orkadiéré and Golf Sud, while in Lambaye this is an idea that is more profound amongst those who themselves know people who have migrated and returned. While stories of migration will arguably be more common to hear in communities where many have migration experience, information about the possibilities of migration may be more restricted to the few who have networks, in places like Lambaye. An important thing to remember when considering the differences between these areas, is that we are referring here to degrees of a culture of migration, and the way of measuring the existence of such a culture is by a relative measurement between the areas. If compared to other areas in Senegal than the ones included here, Lambaye may have high prevalence of networks, and the analysis and findings would have to be interpreted accordingly. As pointed out in the literature review on Senegal in chapter 2, aspects of a culture of migration have been found by qualitative researchers in most areas of Senegal, and thus we may here be speaking only of varying degrees of the same phenomenon.

Based on the results from the logistic regression models in table 7.4, we can say that there are significant differences between the degree to which returned migrant networks are important in Lambaye and Orkadiéré and Lambaye and Darou Mousty. However, we do not know based on table 7.4 alone, whether there are differences between Orkadiéré and Darou Mousty or between Golf Sud and the other areas. Changing reference groups does not yield any other significant results, and thus it is only the difference between Lambaye and these other areas which can be said to be documented in this thesis. Returned migrant networks seem to be significantly more important in the one area where such networks are less prevalent, and perhaps in the one area where a culture of migration is the least present, namely in Lambaye. The fact that there are significant interaction terms with returned migrant networks means that we can say that the degree to which knowing migrants who have returned to Senegal is important for levels of migration aspirations, is something that varies with area. Because there are no significant interaction terms between Lambaye and the other measures of networks, we cannot say that there are significant differences between the areas in these regards.

To conclude this section and the discussion of the independent variables presented in chapter 7.4, we see that migrant networks do play a role in whether or not people want to move abroad. All the coefficients reporting migrant networks are positive, and the coefficients

for knowing both returned migrants and return and current migrants are positive and significant. How much networks matter, depend on the specific emigration environment one is currently living in, indicated by the negative and significant interaction terms. Thus, the first, second and third hypothesis are confirmed, which the exception of people who know only current migrants. The aspects of migrant networks that seem to be important for migration aspirations for people between 18-39 in these specific areas of Senegal, is whether one knows either returned migrants or both return and current migrants.

7.3.3 Control Variables

In addition to the variables on migrant networks and areas, several other determinants of migration aspirations are included in table 7.4. Although included mainly as control variables, a brief discussion of these can inform the picture of what the factors are that shape and influence migration aspirations in Darou Mousty, Lambaye, Golf Sud and Orkadiéré. A presentation of the control variables is also a contribution to the limited scholarly work done on determinants of migration aspirations, some of which were presented in chapter 4 and 5.

When presenting the control variables in chapter 6, I made some assumptions as to how these could be thought to affect the model, and the levels of migration aspirations in the areas surveyed. Previous research on the topic has found that the most important determinants of migration aspirations seem to be age, gender, marital status and employment situation: it is the young, single, unemployed men who are the most likely to want to migrate, according to previous research.

According to the results presented in table 7.4, some of the same important determinants of migration aspirations are found in these four areas of Senegal. Table 7.4 indicates that the most important determinants of migration aspirations, in addition to migrant networks and areas, are marital status and being a student; both substantially increase one’s chance of aspiring to migrate, a finding which is in line with those of Carling (2001) and Van Dalen, Groenewold, and Schoorl (2005). Students are over twice as likely to want to migrate as those who are currently employed, controlled for other determinants of migration aspirations, and having never been married influences aspirations almost as substantially.

None of these coefficients change when the areas are included in model two. These are not surprising findings, as both students and those who are not married are less bound to Senegal when they decide on what they want for their future, and fewer of the restrictions that may apply to migration aspirations, arguably apply to them.

77 Age is highly correlated both with being a student and with having never been married, and this may be one of the reasons why the age coefficient does not become significant before controlling for area in model 2. The coefficient for age is negative and lies close to one, indicating that aspirations are only slightly more common amongst young people than older. This is not as surprising if we take into account that EUMAGINE only included people between 18 and 39 years, which is the age range where most of those who migrate chose to do so. We can assume that other studies have found a larger effect of being young on migration aspirations than can be observed in table 7.4, because their age range is broader.

Each added year of schooling yields a significant increase in migration aspiration. The fact that the coefficient for schooling squared is also significant indicates that the effect is not the same for all levels of education, which is also in line with existing literature.

The wealth index is a measurement of wealth which has been weighted by area. When we observe that the wealth coefficient is negative and significant, this shows that the higher one’s relative level of wealth, the lower one’s aspirations to migrate, controlled for the other determinants of migration aspirations included in the model.

The one determinant of migration aspirations which is not common to include in the literature, but which has been added because of its assumed importance in the Senegalese case, is that of membership in the Mouride brotherhood. This control was included because the Mourides are known to have strong network ties which are of a transnational nature.

Model one in table 7.4 shows that members of the Mouride brotherhood have lower levels of migration aspirations than non-members, but after controlling for area in model two, this effect is no longer significant. The main reason why this effect disappears is probably that what is first observed as an effect of Mouride brotherhood, is explained by the difference between the four areas. The descriptive statistics in table 7.1 show that Mouride brotherhood membership is most common in Darou Mousty and Lambaye. 30 percent of the respondents in Golf Sud say that they belong to this specific Sufi order, while almost none of the respondents in Orkadiéré say the same.

The most surprising finding amongst the control variables is probably that there is no significant difference between the migration aspirations of men and women. Most of the literature finds that men are more likely to aspire to migrate than women, and we also know that Senegalese men migrate in remarkably higher numbers than their female counterparts.

Although not directly linked to this result, the descriptive statistics presented in table 7.1, indicated that women may be overrepresented in the EUMAGINE data material. This is

especially a factor in Lambaye, where only 27.8 percent of the interviewed respondents are men. To see if the gender differences are due to sampling issues, the EUMAGINE team, lacking updated population data, compared the share of women among respondents of the individual questionnaire with the data on all eligible household members. They found the percentage of women amongst our respondents not to be very different from the percentage of women in the surveyed households overall and concluded that the observed gender differences were similar on a household level (Ersanilli 2012). Knowing that there are large differences both in the levels of actual migration, and in the ability to migrate, the finding of no such gender differences seems to strengthen the argument for studying aspirations as a phenomenon separate from ability or actual migration.

To conclude this section, we see that some of the factors known from other studies to be important determinants of migration aspirations are also important for the four Senegalese areas. As indicated also by Van Dalen and Henkens (2008), scholars have had difficulties pin-pointing exactly what factors lie behind such aspirations, and the ones who do seem to vary from context to context, and perhaps also with the ways aspirations are operationalized. The fact that many of the variables included in the model are have values that are close to 1, may be a reflection of the fact that the percentages of respondents who answer that they would go to Europe if provided the relevant papers, i.e. they have migration aspirations, is remarkably high. It may be that we are faced here with a situation where the reasons for why people would chose not to migrate given the situation presented to them in this, are complex and that they are not as systematic. Because almost everyone, man or women, old or young, say that they would move to Europe if provided the relevant papers, it is difficult to find significant and strong effects of for example gender or other things usually find to be important determinants of migration aspirations. In the next section, we will incorporate the insights from the aspiration/ ability model outlined. While levels of migration aspirations are very high in these four areas of Senegal, we know that their ability to migrate is relatively low.

Thus, it will be interesting to see whether the same determinants of migrations aspirations are valid when we discuss migration intentions rather than migration aspirations. That is the topic of the next section.

79